How to track your AI search visibility
Why you have to track it, not guess
AI answers are not static. Engines re-crawl the web, refresh their indexes and update their models, so the company an engine names for "best {your category}" this month may not be the one it named last month. That means AI visibility is a moving position, not a one-time verdict — and the only way to know whether your work is paying off is to measure the same thing repeatedly. A single spot-check tells you where you stand today; a tracked trend tells you whether you are winning or losing ground, and whether a specific change actually moved the needle.
What to measure
"Am I in ChatGPT?" is too blunt to manage against. Break it into inputs you can record for every prompt, on every engine, every time:
| Metric | What it captures |
|---|---|
| Cited (yes/no) | Were you named or linked at all for this prompt on this engine? |
| Share of voice | Out of every company the answer named, what fraction was you across the whole prompt set? |
| Prominence | First-named and linked counts more than a passing mention at the bottom |
| Who's cited instead | The competitors and sources winning the citation — your roadmap of what to match |
| Engine spread | Which engines cite you and which skip you — visibility is rarely even across all four |
How to track it — step by step
Lock a prompt set that reflects real buyers
Write down ten to fifteen questions your buyers actually ask an AI assistant — "best {category} for {buyer}", "who should I hire to {job}", "is {your company} any good", "alternatives to {competitor}". These become your fixed benchmark. Resist the urge to change them between runs: the moment the questions drift, the trend stops meaning anything. Add new prompts only as a separate, clearly-labeled set so your core trend line stays comparable.
Run every prompt through every engine, the same way
Ask each prompt in ChatGPT, Perplexity, Google AI and Claude. Use fresh sessions so prior chat history doesn't bias the answer, and keep the wording identical each month. Consistency of method is what separates a real measurement from an anecdote — same prompts, same engines, same conditions, so the only thing changing between runs is the web itself.
Record the result for each cell
For every prompt-and-engine pair, log whether you were cited, who else was named, and how prominently you appeared. A simple spreadsheet with a row per prompt and a column per engine is enough to start. The "who's cited instead" column is the most valuable thing on the sheet — it tells you exactly which pages and sources are beating you, which is your to-do list.
Roll it up into one score
Collapse the grid into a single 0-to-100 visibility score — a share-of-voice number for how often, and how prominently, you show up across the whole set. The exact formula matters far less than keeping it identical every month. One comparable number per month is what turns scattered observations into a trend a decision-maker can act on.
Re-run on a fixed cadence and read the trend
Repeat monthly. Watch the score line and, just as importantly, the composition underneath it — a flat score can hide you gaining ground on Perplexity while slipping on Google AI. When the number moves, tie it back to what you changed: a new answer page, a fixed crawler block, a fresh third-party mention. That cause-and-effect is the entire point of tracking.
How often to check
Monthly is the right default. Engines don't re-crawl and re-rank fast enough for daily checks to show anything but noise, and chasing day-to-day wobble wastes effort. The one exception: right after a meaningful change — you unblocked GPTBot, shipped a new answer page, or cleaned up your entity — a follow-up check in a couple of weeks confirms the engines have re-fetched you and the change is landing. Otherwise, once a month keeps the signal clean.
Doing it manually vs. monitored
You can absolutely track this yourself for free with a spreadsheet and an hour a month — and if you only care about a handful of prompts, you should. The reasons businesses hand it off are consistency and history: running the exact same prompts the exact same way every month, never skipping a run, and keeping a clean multi-month record you can put in front of a boss or a board. That is what a monitoring program buys — disciplined measurement over time, not a promise about the outcome. Be wary of anyone selling a "guaranteed score" instead of honest measurement; a guarantee is the #1 red flag.
What tracking can't do
Measurement tells you where you stand and whether you're moving — it doesn't move you. The score going up is the result of fixing the underlying inputs: crawlability, indexation, answer-ready pages, a consistent entity and third-party corroboration. Tracking simply makes that work accountable. And no honest tracker will promise a specific number by a specific date, because what an engine cites depends on the whole web, not just on you.
Frequently asked questions
How do I track whether AI mentions my business?
Fix a set of buyer-intent prompts, run each through ChatGPT, Perplexity, Google AI and Claude on a regular cadence, and record for every prompt whether you were cited, who was cited instead, and how prominently. Roll it into one share-of-voice score and repeat on the same prompts each month so the trend is comparable.
What is an AI visibility score?
A single 0-to-100 number summarizing how often and how prominently AI engines cite you across a fixed prompt set — a share-of-voice measure. The formula matters less than keeping it identical every time you measure.
How often should I re-test my AI visibility?
Monthly for most businesses. Engines re-crawl and update over weeks, so daily checks mostly capture noise. Check sooner only right after a major site or entity change, to confirm engines re-fetched you.
Can I track AI visibility for free?
Yes, manually — pick 10-15 prompts, ask each engine, and log the results in a spreadsheet. The risk is inconsistency, which produces a trend you can't trust. Tools and managed monitoring mainly buy consistent, scheduled testing and clean history — not a guaranteed score.
See your score before you start tracking it
Our free AI Visibility Audit runs your buyers' real questions through ChatGPT, Perplexity, Google AI and Claude — and gives you the baseline score, the prompt set, and who's cited in your place, so you have something to measure against.
Request the free auditRankInAnswers measures the inputs that drive AI citations and reports the trend every month. We never guarantee what an AI engine will say about you — no honest provider can.